Documentation

LeanPool.PebblingLean.LowerBound

Lower-bound infrastructure for hypercubes #

This file starts the formal version of the weight-function lower bound. The main result here is the first inequality in the paper's lower-bound proof: for a solvable distribution on Q_n, every target has initial weight at least one, and hence the sum of all target weights is at least |Q_n|.

For the hypercube, any target reachable from D has initial weight at least one with respect to that target.

A solvable distribution has target weight at least one for every target.

Summing the target-wise lower bound over all targets gives |Q_n| ≤ ∑_t W_t(D).

theorem PebblingLean.Hypercube.sum_invPow_card_finsets (n : ) :
s : Finset (Fin n), 1 / 2 ^ s.card = (3 / 2) ^ n

The finite-subset identity behind the inner hypercube weight sum.

theorem PebblingLean.Hypercube.sum_unitWeight (n : ) (base : HypercubeVertex n) :
target : HypercubeVertex n, Pebbling.unitWeight dist target base = (3 / 2) ^ n

For a fixed hypercube vertex base, the sum of one-pebble weights over all targets is (3/2)^n.

theorem PebblingLean.Hypercube.sum_weight_eq_size_mul {n : } (D : Pebbling (HypercubeVertex n)) :
target : HypercubeVertex n, Pebbling.weight dist D target = D.size * (3 / 2) ^ n

The double sum of weights factors as |D| * (3/2)^n.

Formal version of the averaged weight inequality |Q_n| ≤ |D| (3/2)^n for every solvable distribution D.

The lower bound in the paper's usual rational form: every solvable distribution on Q_n has at least (4/3)^n pebbles.

Lower bound for any specified solvable size.

Lower bound for any k certified as the optimal pebbling number.